INDEX

Explanations

programming-related concepts and syntactical structures

oai_token-act-pair · gpt-4o-mini Triggered by @bot

calibration, user, string

np_acts-logits-general · gemini-2.5-flash-lite

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B-IT @ 20-gemmascope-res-131k

Configuration

google/gemma-scope-9b-it-res/layer_20/width_131k/average_l0_81

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

Data Type

float32

Hook Name

blocks.20.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Durchmesser

-0.38

 linho

-0.34

 syscall

-0.32

 another

-0.31

Disposable

-0.31

 Verfügung

-0.30

 Verfü

-0.30

 fevere

-0.30

大陸

-0.30

 Maurer

-0.29

POSITIVE LOGITS

setVerticalGroup

0.73

Jeografia

0.58

 <<<<<<<<<<<<<<

0.58

findpost

0.56

发表于

0.56

帖最后由

0.53

Chham

0.51

 تضيفلها

0.50

CppMethod

0.50

LabelTagHelper

0.49

Activations Density 0.008%

programming-related concepts and syntactical structures

calibration, user, string

No Comments

No Known Activations